Hi Guys
MJE15034/35 look a lot like MJE15030/31 except for having a much higher voltage rating. Gain is fairly flat and consistent over normal driver-stage current requirements. GBW curves for the 34/35 look a little scary for low currents, but the same curves for 30/31 do not go below 100mA, so may look similar.
The lower-voltage MJE15030/31 is ubiquitous and has proven itself to be a favoured choice in many high-end designs. Bryston has used them since they became available. The 30/31 are seen in amps up to +/-75V rails, even though their rating is only 150V. If you are doing a class-G amp or something with +/-100V rails, the 34/35 should give similar performance.
Have fun
Kevin O'Connor
MJE15034/35 look a lot like MJE15030/31 except for having a much higher voltage rating. Gain is fairly flat and consistent over normal driver-stage current requirements. GBW curves for the 34/35 look a little scary for low currents, but the same curves for 30/31 do not go below 100mA, so may look similar.
The lower-voltage MJE15030/31 is ubiquitous and has proven itself to be a favoured choice in many high-end designs. Bryston has used them since they became available. The 30/31 are seen in amps up to +/-75V rails, even though their rating is only 150V. If you are doing a class-G amp or something with +/-100V rails, the 34/35 should give similar performance.
Have fun
Kevin O'Connor
Still playing around
Ditched the CCSs to try bootstrapped resistive loading and it isn't too shabby. It definately isn't as touchy with getting offset in check. I think I'd like to play with the BIGBT some more, maybe just go stupid and try the BJTs in a CFP driven by a smaller MOSFET. Simmed distortion doesnt change much until just shy of clipping, about 3V off the rails.
Ditched the CCSs to try bootstrapped resistive loading and it isn't too shabby. It definately isn't as touchy with getting offset in check. I think I'd like to play with the BIGBT some more, maybe just go stupid and try the BJTs in a CFP driven by a smaller MOSFET. Simmed distortion doesnt change much until just shy of clipping, about 3V off the rails.
Attachments
No, I just used the models that best fit what I currently have on hand, which is some 2SC3281 and 2SA1302 I bought in the early nineties and never used. It wasn't my intention to confuse the issue showing specific devices.
Hi Guys
Jkuetemann, see my earlier posts regarding non-CS versions of this buffer:
114 - Rs for CSs
116 - schemo showing Rs for CSs and class-A bias servo
136 - paragraph-5, discusses offset advantges of Rs
186 - performance of bootstrapped-R version with triple-EF, doubled outputs, local filters, plus layout and schemo - THD <0.002% measured across the band (limit of Radford test system)
All my circuits have proper input and output referencing, as detailed earlier.
In the past I tended to add CSs and current mirrors to assure performance and to eliminate critical R values in certain circuits. After building a lot of different things and many of each, I'm finding certain simplifications to be worth the added effort of matching devices. The overall distortion profile becomes balanced, benign and low-order.
Have fun
Kevin O'Connor
Jkuetemann, see my earlier posts regarding non-CS versions of this buffer:
114 - Rs for CSs
116 - schemo showing Rs for CSs and class-A bias servo
136 - paragraph-5, discusses offset advantges of Rs
186 - performance of bootstrapped-R version with triple-EF, doubled outputs, local filters, plus layout and schemo - THD <0.002% measured across the band (limit of Radford test system)
All my circuits have proper input and output referencing, as detailed earlier.
In the past I tended to add CSs and current mirrors to assure performance and to eliminate critical R values in certain circuits. After building a lot of different things and many of each, I'm finding certain simplifications to be worth the added effort of matching devices. The overall distortion profile becomes balanced, benign and low-order.
Have fun
Kevin O'Connor
Way to go
I can not tell you how pleased I'm with them and how good these are in real world practice. Only that I prefer to use verticals (IRF610/9610) for drivers as they're faster and likes to drive more current. VAS stage, no matter which topology, driving BIGBTs breathes with full lungs at all times.
Not correct, Typical ft at 30ma is 150Mhz. The most important factor is the 30ma, which is where a typical driver is biased at. At this current toshibas manage around 70Mhz and onsemi a dismall 20 to 30 Mhz.
No thanks, I dont want what will probably be fakes or junk second source. The bds have long been discontinued and only the fairchild second source seems to be capable of similar performance. Clearly you dont understand the distortion mechanisms with push pull emitter followers so I wont waste further time with you on the subject in this time zone.
Hi Guys
Clearly I do not understand things the same way as you, homemodder. Rather, what I understand is that there is more to BJT performance than mere Ft. Gain flatness over the current range actually used is more important and does not require that this be coincident with the maximum Ft point of the device. In this view, I am joined by name designers whom I respect.
Lazycat, I finally took a closer look at your discrete IGBT circuit and understand why you like it. If you took out the diode between the BJT collectors, you have a formof Bryston Quad-complimentary output stage in effect, with a mosfet driver instead of BJTs, and without voltage gain. The combination of an emitter and a collector working the load for each signal half has a profound effect on the distortion spectrum, generally making it more balanced and benign.
Have fun
Kevin O'Connor
Clearly I do not understand things the same way as you, homemodder. Rather, what I understand is that there is more to BJT performance than mere Ft. Gain flatness over the current range actually used is more important and does not require that this be coincident with the maximum Ft point of the device. In this view, I am joined by name designers whom I respect.
Lazycat, I finally took a closer look at your discrete IGBT circuit and understand why you like it. If you took out the diode between the BJT collectors, you have a formof Bryston Quad-complimentary output stage in effect, with a mosfet driver instead of BJTs, and without voltage gain. The combination of an emitter and a collector working the load for each signal half has a profound effect on the distortion spectrum, generally making it more balanced and benign.
Have fun
Kevin O'Connor
Last edited:
Not more important but as important as is the cob and the early voltage of the device / Low saturation voltage comes to mind in second plane. Gain flatness with maximum Ft is not required but if you want high performance both can be achieved at the same time with clever device selection. Try upping the bias on the driver on any class AB amp and youll hear the difference, especially with the dismall onsemi devices. These should are really only be used with triple EF stages and very high power amps were other devices are unfortuneatly not suitable or exist.
The original Bds have a gain which is just as flat as any toshiba or MJL, as a bonus you get higher beta which is advantageous and you get much improved Ft which is also advantageous. Thirdly the cob is also lower meaning lower high frequency distortion.
The BDs from fairchild are not far off the originals or mje243 and mje253.
Higher voltage youd have to use toshibas, parralel them if you have to, even then much better than the onsemi devices. Another current ex sanyo and now onsemi is 2sa1507 and complement which is comparable to the best toshiba pair 2sa1930 and compliment.
Higher voltage youd have to use toshibas, parralel them if you have to, even then much better than the onsemi devices. Another current ex sanyo and now onsemi is 2sa1507 and complement which is comparable to the best toshiba pair 2sa1930 and compliment.
Hi Guys
Saturation voltage is important in a switching circuit when you want low losses, but seems irrelevant in an audio amp. I don't run my amps to clipping, nor do I need to extract that last 100mV of swing from the supply rails.
Jkuetemann, the mosfets in your sim are now obsolete according to Toshiba. However, the K1058-J162 may still be current. They are all close enough to each other in performance to be swappable for the circuit. It is disconcerting to see these obsoletions occur, especially when you have these parts in your inventory; should you use them? keep some as spares?
Fortunately with the plastic packages, the mosfet and BJT pinouts are compatible. So, if you lay out a board for the LU buffer, you can use either technology. Just make a provision for the extra drive or predrive stages, and for gate zeners and balancing caps. Then, if the original mosfets die and cannot be replaced, BJTs can be subbed, or vice versa.
Have fun
Kevin O'Connor
Saturation voltage is important in a switching circuit when you want low losses, but seems irrelevant in an audio amp. I don't run my amps to clipping, nor do I need to extract that last 100mV of swing from the supply rails.
Jkuetemann, the mosfets in your sim are now obsolete according to Toshiba. However, the K1058-J162 may still be current. They are all close enough to each other in performance to be swappable for the circuit. It is disconcerting to see these obsoletions occur, especially when you have these parts in your inventory; should you use them? keep some as spares?
Fortunately with the plastic packages, the mosfet and BJT pinouts are compatible. So, if you lay out a board for the LU buffer, you can use either technology. Just make a provision for the extra drive or predrive stages, and for gate zeners and balancing caps. Then, if the original mosfets die and cannot be replaced, BJTs can be subbed, or vice versa.
Have fun
Kevin O'Connor
What I have found is that the addition of the inductor makes certain ringing into a capacitive load. Can you hear that? I dunno... Thinking out loud now, I wonder if anyone has done actual distortion measurements with the ringing?
_-_-bear
Member
Joined 2009
Paid Member
I think ST might still be doing these, but not sure of the spec.
They were nice GP devices in their day, but a bit let down by the low Vceo rating.
I have Philips handbook fro 1974 and it states fT for BD 135/137/139 is 250 MHz anf fT for BD136/138/140 is 75 MHz nothing about Cob.
dado
The original philips datasheets state ft of 160Mhz for pnp and 190 for npn but please note that these were taken at vce of 5 volt where nowadays its mostly taken at 10v which makes things look a little better but I believe the ft of the npn part could well reach 250 Mhz at vce of 10v. Thats how good these parts are, a real pity about the max Vce but you could push a paralelled pair to +- 50 volts if you re adventurous. Max Vce is stated open base.
The rest of the info can be had looking at datasheets for bc638/9 ect. The bds are simply these in to126 package, this is what the philips parts are and why the spice parameters are identical for the parts. The onsemi ones however are nowhere close to the originals so I dont even consider them as being bd components.
I agree better vas transistors are available but at least they are much better than mje340/350 which are commonly still found even on commercial products.
The rest of the info can be had looking at datasheets for bc638/9 ect. The bds are simply these in to126 package, this is what the philips parts are and why the spice parameters are identical for the parts. The onsemi ones however are nowhere close to the originals so I dont even consider them as being bd components.
I agree better vas transistors are available but at least they are much better than mje340/350 which are commonly still found even on commercial products.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.